Perspective Switching in Virtual Environments

When exploring new environments, people regularly alternate among many sources of spatial information including direct visual input, navigation aids such as maps and mobile devices, and verbal route descriptions. These spatial representations typically depict the environment from one of two perspectives: first-person, embedded route perspective or top-down, bird's eye survey perspective. Visual spatial cognition research has explored the nature of learning within each of these perspectives independently, but little work has been done to explore how on-line visual processing of combined perspectives affects cognition, meaning there is little understanding of the cognitive costs of using different navigation tools to learn large-scale environments. This dissertation addresses such questions through two experiments that guide participants through simple paths in large-scale environments, each consisting of a simple path through a small virtual town presented on a desktop computer display. By timing participants' movement through each environment and how they respond to either externally-controlled or participant-controlled perspective switches, the experiments measure the cognitive load of visually processing dynamic perspectives during navigation. These on-line processing measures are complemented by tests of visual recognition and recall memory, which reveal how switching perspectives affects the accuracy of the resulting spatial mental model. The results indicate that the cognitive load associated with changing perspectives is primarily dependent on the quantity of visual information the change introduces -- the transformation itself is not particularly disorienting after the first exposure to the environment. Furthermore, although forced perspective switches do not appear to significantly affect spatial memory accuracy relative to viewing the environment from a consistent perspective, navigator-controlled switching results in significantly more accurate spatial memory, indicating that navigation aids which allow for perspective control might better support spatial learning than fixed-perspective interfaces. The findings also support previous research showing that route perspective navigation generally yields more accurate spatial memory than survey perspective learning, particularly after extensive experience in the environment. Overall, the findings demonstrate many new aspects of how perspective affects spatial cognition, with implications for spatial learning and the design of navigation aids.